Weird manifesto for UK ARPA or ARIA

My ideas:

-Progress Studies (including social progress and creativity)

-Basic Climate research

-How creativity happens

-Productivity schedules (sleep, diet, schedules)

-Educational Mastery

-Building Speed (how to do big projects fast[er])

-Healthcare speed, innovation, public health challenge trials,

The UK is creating a £800m sciency agency based on the US ARPA* an innovation agency. The UK agency will be called ARIA, Advanced Research and Invention Agency. This idea had considerable backing from former special adviser Dominic Cummings (see his lengthy blogs on this, links end*). While it has received criticism and isn’t a novel idea in innovation circles, we have it. So let’s make the most of it.

What is ARPA?

In brief at ARPA, around 100 program managers (PMs) with ~5 year appointments create and run programs to pursue high-level visions, for instance - what would make electric planes or hydrogen heating systems.

There is much written on what makes US DARPA work with a decent focus on having brilliant PMs. Also the similarities and differences on how IARPA (intelligence focused) and ARPA-E (energy focused) have also worked. So, I won’t dwell on that. 

UK’s ARIA is likely to be given the freedom to choose what it works on. Let’s put aside the debates on ARIA and the long history of innovation policy experts here and posit what we think ARIA should work on.

I’d like to float some weird and not so weird ideas I think a UK ARIA should focus on. I have a touch of weirdness about me so that fully qualifies me (tongue in cheek).

These ideas support areas which would have large public goods benefits (and some private sector benefits) but for various incentive/time horizon problems are not well suited to private actors.

I have 3 large buckets but with some off sub-buckets.

•  Basic Climate research

•  Progress Studies (after Cowen & Collinson)

•  Healthcare (life extension and quality of life extension)

Basic Climate research: Trees and Seas 

While this falls under “net zero” area, my idea under basic climate research is more foundational than eg tackling hydrogen based systems for making carbon neutral steel.

There are several areas here, where - it seems to me - we simply do not understand the state of the world and its system but we might now have the technology and research to do this.

For instance, what is the true state of our forests, jungles and trees over the globe? Data and the interpretation of that data is unclear. Where are trees disappearing, where are we planting and how is it going?

Bill Gates is dismissive of trees as a climate solution*. The UN FAO has data, visualised by World in Data*, but other attempts to assess trees are contradictory.

This is due to problems of classifying types of trees (shrubs, types of trees etc.) and the aerial data needed. And there are problems with losing old trees (especially primary rain forest) and replacement of new trees.

I am far away from the literature and no expert but I sense a programme here and maybe one specific to UK and UK peat lands, tree afforestation etc. would be very useful in basic research.  Essentially, the same argument for seas and oceans and their contribution to carbon sinks.

Basic climate research is not super weird. But I think there are big basic knowledge gaps here which could be very valuable and items like trees are not best suited to private actors.

Progress Studies

• How innovation happens, how to make it better. Same for social and ethical progress. Also,

• Creativity, flow, educational mastery

This is weirder although Tyler Cowen, Patrick Collinson et al is making it a lot less weird.

This would need to go beyond “Cities and innovation clusters show agglomeration affects” (known and somewhat trite, IMO, in that difficult for policy to seeming build upon) but can we drive real insights here? Small teams? Big teams? Collaboration from cross-disciplines which are neither too far, nor too close. The impacts from regulation (are the de-regulation cries correct in all respects? ). 

There will be a tendency to look at this in the hard sciences and the inventions, innovations etc. there (and there is a literature here). And I think understanding that will be useful especially eg in medical science, software and the like, but my weird question is what about social progress?

There is consensus today that slavery is bad. I think that counts as social progress. But how does that happen? What role does “culture” play?

I think society is increasingly valuing eg autistic thinking and (while there is much further to go), we have given some more rights and some more status to the spectrum of autistic thinking and other areas like this.

I think some rigorous work here would be insightful and useful. If the productivity or progress can be raised in these areas there could be strong benefits.

Running along side this, I’d be interested in rigorous work on Team and Individual productivity progress.

There could be an enormous win if robust findings could be confirmed here.

For example, Paul Graham (extremely successful in the start-up founder and investment space) has argued that maker time and manager time are very different schedules.

Essentially, maker time requires good lengths of the day devoted to the creative projects  (in my view, related to what we know about flow) whereas manager time needs shorter chunks of meeting time.

Where manager time interferes with maker time, you get a huge negative impact to maker productivity.

If this is correct and if we can guide for it, this could improve productivity and be of general benefit. Would this be progress? I think so, and of general public good. Why are there so many time management books? Tyler Cowen amongst others often asks about people’s “personal productivity function” ?  Can we actually discover anything robust here?

Let’s go one step further, we seem to have some tentative ideas about sleep and productive circadian rhythms of the day for certain people (eg night owls).

We have tentative ideas about intermittent fasting or diet and potential health benefits.

Is there any work on trying to combine these factors or ideas? 

If you current have poor productivity, but what you should do is change to a nightowl, maker schedule on an intermittent fasting schedule - could your productivity significantly increase?

And then how about combining this within teams? There is work on psychological safety*, and some thoughts as to  innovation seems to happen when teams understand each other’s work but are not too close or not too far away - but can we combine any of these possible insights?

From this can we create even more builder teams, like the Tesla’s, Stripes., etc of the world.

Perhaps it is too abstract and too difficult to do rigorously, but I think this would be a weirdly good area for UK ARPA to examine.

As extension, I’d look in to how we foster creativity. Specifically, I’d be interested in extending the work around “flow” and any rigorous study on the structure of “story” or “narratives”. And also an examination of forms of “educational mastery” 

Flow

Why flow? There is some suggestion that flow can significantly increase creative productivity (although there might be downsides in using flow to enter practice states that don’t lead to new development). Rigourous work around here that might be more widely applied could have strong benefits. Same for overall creativity.

Mihaly Csikszentmihalyi (Creativity: Flow and the Psychology of Discovery and Invention, 2013) has done work here but can it be extended and made wider known? If it can raise the creativeness of our top 20% (or anyone) could there be huge gains?

Drama and story

The basis of most (western) dramatic structure was written by Aristotle in his poetics around 330 BCE - so over 2300 years ago. While we have had some incremental changes and Shakespeare arguably stepped up this form there are a couple of way of thinking about this. One is that drama and story has been stagnating for a long time but another is that there is something fundamental about story structure that has persisted over centuries (maybe something Lindy? As Taleb might say)

Given the way that story/narrative/myth seems to really impact human behaviour (intersubjective myth for instance) and in world where humans might benefit being resistant to mis-information - I think there could be good gains from a rigourous study here.

Thinkers like Ray Dalio put strong weight on the hero quest story arcs in life and my weird suggestion is that a study around what we know about “story” as a social science exercise  would be insightful. 

This is probably too leftfield for them, but my next idea could be more mainstream and that is an examination of “educational mastery” especially in the context of online or Khan academy type innovations.

Educational Mastery

Patrick Collinson writes: “Educational psychologist Benjamin Bloom found that one-on-one tutoring using mastery learning led to a two sigma(!) improvement in student performance. The results were replicated. He asks in his paper that identified the "2 Sigma Problem": how do we achieve these results in conditions more practical (i.e., more scalable) than one-to-one tutoring?

In a related vein, this large-scale meta-analysis shows large (>0.5 Cohen's d) effects from direct instruction using mastery learning.

Is this a true effect and can we do more about it? Can it scale using online methods? 1-1 video ? Or if not, is there value to eg. randomly (or not) selecting some students and giving them mastery type learning. If just these small groups have two sigma improvements - could we see some significant gains?

I think ARPA could well study something in this area.  Nintil* did a thorough research round up suggesting the Bloom effect was not as large. But, 1-1 teaching did have a very robust effect. 

We could find a number of people willing to give 1-1 teaching as extra and maybe a number of students (across high performing or medium/low performing groups). If 1-1 can dramatically improve performance would this be worth studying or working on?

Building faster

Lastly, in this area it would be useful to examine why we seemed to be able to build infrastructure and certain other items faster 50 years ago. First, how true is this? UK managed to build Olympic sites in a moderately fast time frame but not eg. the tube extensions. This might not exactly be an ARPA area, but I think it could under pin a lot of innovation. (cf again Cowen, Collinson).

I think there’s an enormous amount that we do better, but can we learn from where we had speed before. Are there robust findings here? Or it just a nice to think venture capital thing.

My last huge area is on healthcare. 

Healthcare progress

I would also suggest there is work done on studying healthcare progress. Now there is a huge literature here, but I see less in a cross-disciplinary nature. This is intersectional with some other ideas here, but it would be what discoveries have most improved human health and how can we have more of them? What are the barriers or not.

Hand washing, weight control, diet, exercise and other low cost interventions are known but how best to synthesise this and can it be combined with newer technolofy and how intersectional with the social determinats of health?

This area will be a focus areas coming off the pandemic, but there is - to my reading - limited work on synthesing how best human health can be improved and the barriers to it.

And this is because of the incentives of where the private sector will focus its innovation and capture public good improvements or not. 

There are potentially very strong and perhaps moderately easy wins here. Two areas would be cost/benefits of areas of drug regulation. The UK has a particular opportunity here.

For instance, it could use EMEA and US  regulatory equivalence but go further and decide to approve certain medications quicker than those regulators. ( I think patient choice could be interesting here, post phase II and/or safety studies)

The UK could extend ideas it has started on “challenge trials” to see if this could significantly speed up areas of therapy development. There are areas probably more areas suitable for challenge trials and areas less suitable and not only COVID. ARIA could run a programme assessing and potentially funding some of this.  

Where would the cost/benefits of challenge trials help the UK/World in certain disease areas?

ARIA could go beyond narrow areas of regulation and even challenge trials but try an synthesis areas of public health.

Can robust work be done on how eg digital health data combined with preventative interventions could make huge, inexpensive, health interventions.   I think this could be a huge area. Many pilot trials have started (eg see a lot of the work Optum do) but some rigorous programmes here could be of enormous value.

In sum, we have ARIA. Let it explore some weird ideas. A few more transformational weird ideas would be a good thing and won’t displace all the other R&D things we are doing.

Links:

Dominic Cummings blog

On trees, World in Data but here on the conflicts in the data and conflicting data here.

Paul Graham, maker time

On Flow: Mihaly Csikszentmihalyi. Creativity: Flow and the Psychology of Discovery and Invention, 2013

Nintil on educational mastery

Patrick Collinson, fast things. And Cowen and Collinson on Progress Studies. 

Policy Exchange: https://policyexchange.org.uk/wp-content/uploads/Visions-of-Arpa.pdf

ON ARPA https://benjaminreinhardt.com/wddw